一种改进的RNS算法的研究及应用-河北科技大学学报.PDFVIP

32 3 Vol. 32, No . 3 20 11 6 Journal of Hebei University of Science and T echnolo y June 20 11 : 1008- 1542( 2011) 03-0256- 05 RNS 王宏宇1, 2 ( 1. 华东理工大学信息工程与科学学院, 上海 200237; 2. 石家庄职业技术学院信息工程系, 河北石 家庄 050081) : 针对异常检测问 , 改进了混合神经元免疫系统中实数阴性选择算法( RNS, real-valued ne ative selection) , 使其更好地应用在多种纬度的空间上实验结果表明, 使用改进后的实数阴 性选择算法( GRNS, eneral RNS) 的神经元免疫系统仅用很少的自体集, 就可以达到较高的检测 率和比较低的误警率, 在高维度特征量的异常检测中取得了很好的效果 : 混合神经元免疫系统; 阴性选择; 神经网络 : T P393. 08 : A Study and application of an improved RNS al orithm 1, 2 WANG Hon -yu ( 1. School of Information Science and En ineerin , East China University of Science and T echnolo y , Shan hai 200237, China; 2. Department of Informat ion En ineerin , Shijiazhuan Vocational T echnolo y Institute, Shijiazhuan Hebei 050081, China) Abstract: For anomalous detect ion, w e improved t he rea-l valued ne ative selection ( RNS) al orithm of hybrid neuro-immune sy stem ( H NIS) , and named it eneral RNS ( GRNS) . The al orithm is applied to neurons-immune system for anomalous detec- t ion. T he results show that, w ith such small number of self sam ples, the improved NIS has hi h detect ion rates and low false alarm rates. Key words: hybrid neuro-immune sy st em; ne ative select ion; neural netw ork ( artificial immune sy stems, AIS) , , , [ 1] ( hybrid neuro-immune sy stem, HNIS) ( r ea-l valued ne ative selection, RNS) HNIS , , RNS , , ( GRNS, eneral RNS) , 1 ( HNIS) HNIS , , 1. 1 ( RNS) RNS , M ILA RN S : 2011-01-22; : 2011- 03-08; : : ( 1969- ) , , , , , 3 RNS 257 1. 2 ( mult-i layer perceptron ML